The chorus of people been calling for a pause has gotten louder. I'm ready to join them. Until recently, generative AI was more of a promising novelty than a useful tool. Work theater rather than work. Sometime over the last 6 months this changed. LLMs can do real work, enough that workers and companies and governments need time to catch up to the present state of things.

The rare voice of reason. Which means it will be drowned here in minutes, and probably get flagged.

> They are built to make VCs and companies rich.

That, and the concomitant terror of incessant breathless shilling and astroturfing. These wankers make so much noise, I can't hear myself think. Which of course is on purpose. Once I get to thinking, I might not buy what they want me to.

> This isn’t sustainable.

I so hope this is true. But the market can stay drowned in propaganda longer than my patience lasts.

the tragedy in this is the people who control the money - VCs etc - and people who ask for money are all from the same backgrounds - typically white & Ivy educated. Groupthink.

hence it's a vicious self-deluding bubble that has massive reinforcement. likewise the talent/capital that could be going to useful impactful opportunities in society doesn't.

hence we get money & labor being pumped into non-problems and solutions seeking problems like LLMs and the massive propaganda going into it as if the propaganda is going to cause behavior change.

which is why diversity of thought, experiences, culture is very important.

I think this and Lopatto's article are insightful and worth repeating... But they're missing the mark on genAI. GenAI is most certainly something wanted by the heavily-moneyed customers of the surveillance-spam-slop industry (aka Silicon Valley). It's not just governments who have the money to pay for it, rather businesses do as well. And this gets right back to the concept of "what normal people want" - not as a static thing, but rather as something that can be shaped and molded.

> Generative AI has no shortage of ways that it might, with care, be shaped into genuinely useful products, but this shaping needs to actually happen before the hyper-scalers earn the right to continually harass the psyche of billions of people with breathless pronouncements. Most people don’t care that GPT 5.5, released late last week, underperformed Opus 4.7 on SWE-Bench Pro. They want the AI companies to let them know when they have a product that will actually and notably improve their lives, and until then, they want these companies to leave them alone and try their best not to crash the economy .

I don’t want to be snarky or dismiss without giving it though but this line of thinking is childish and shows lack of understanding of how products enter market. Companies that invent product generally can not predict all the uses. That’s what the free market exists for - push your product and see how people use them, get feedback and improve. I find that beautiful.

Does the author want planned RCTs to understand how they might be useful and slowly release them only every dimension of utility is confirmed? This reeks of euopilled mindset - a perverse parental kind of mindset where the state decides how products may be used.

The Silicon Valley doesn’t work like this. They release products and the second or third order effects can not be predicted. Who knew NVIDIA might be used for LLMs when they first started producing chips for gaming?

The author (I assume) is feeling some latent anxiety over the extreme speed in which things are progressing. They are grieving by finding reasons to stall this progress by falsely claiming that people don’t care about LLMs until there is confirmed proof that LLMs can help them improve their lives.

But this is fundamentally a childish mindset, and flat out wrong. People do care - there are always enthusiasts who are top of the game and who the broader public follow.